Autoregressive World Model
A generative model that takes the current or historical state as input and predicts future states step by step, typically treating the scene representation as a token sequence and reasoning in temporal order with structures such as Transformers.
Timeline (last 90 days)
GeoState contains three key components: a scene tokenizer, a geometry decoder and an autoregressive world model; the scene tokenizer aggregates multi-view GFM features into a viewpoint-decoupled scene state; the geometry decoder queries the scene state with camera rays to render observations from a specific viewpoint; the autoregressive world model performs prediction in the decoupled scene state space and models camera motion independently.
An autoregressive world model refers to a generative model that takes the current or historical state as input and predicts future states step by step, commonly treating the scene representation at each moment as a token sequence and performing inference in temporal order with structures such as Transformers.
All Facts (2)
GeoState contains three key components: a scene tokenizer, a geometry decoder and an autoregressive world model; the scene tokenizer aggregates multi-view GFM features into a viewpoint-decoupled scene state; the geometry decoder queries the scene state with camera rays to render observations from a specific viewpoint; the autoregressive world model performs prediction in the decoupled scene state space and models camera motion independently.
50%UnverifiedAn autoregressive world model refers to a generative model that takes the current or historical state as input and predicts future states step by step, commonly treating the scene representation at each moment as a token sequence and performing inference in temporal order with structures such as Transformers.
50%